TerraByte is a technical operation focused on AI infrastructure, web systems, and software that ships.
The work started in web development — 25+ client sites delivered across restaurants, contractors, fitness, and professional services. That foundation gave way to something more technically interesting: production AI systems. Today, 30+ AI agents are running across 6 production deployments on Linux, Mac, and Windows — handling real workloads for real businesses, not demos sitting in a sandbox.
The current stack spans TypeScript, React, Rust, and Python depending on what the problem requires. On the AI side: multi-tenant agent deployments with isolated knowledge bases, RAG pipelines over proprietary business content, streaming LLM integration across multiple providers (Anthropic, OpenAI, Ollama), and agentic systems built around tool use and persistent memory.
Beyond client work, active development is happening on desktop application software using Tauri v2 — native cross-platform apps with Rust backends and React frontends, integrating directly with LLM providers at the protocol level. This is where most of the deep technical work happens: streaming inference, context window management, multi-provider abstraction, local model support via Ollama.
The throughline is that this work is genuinely interesting. LLM architecture, inference optimization, agentic system design, prompt engineering at depth — these are studied because the field moves fast and there's no substitute for staying close to it. Every system running in production reflects that: built on understanding the fundamentals, not copying a tutorial.